Re: [condor-users] Windows Condor disappearing/reappearing on gri d

[Antti Eskola <antti.eskola@xxxxxxx>]

On 15 Nov 2003 11:19:01 +0200 Mark Silberstein wrote:

Does all workstations in the pool have clock at the same time (inside 5
minutes)?...
I have had big job trying to get a solaris workstation to my pool (it has 
linux and windows at the moment they work ok)
the solaris stations have clocks set to time that differs alot from the
rest of the stations.?
Tomorrow i try to install new release 6.6 to all nodes and hope it will
solve some problems.

when i submit a job defined for solaris station it will be matched and
best time the job started but then it says cannot longer talk to the
solaris shadow process... (no firewall between Central manager and
solaris station)

if i cant succeed on this next test phase i will give you m8ths all the
information i can (logs, etc...) about the submission because i am all
out of ideas at the moment...

And also i usually get when trying to het this work messages in the
solaris logs that permission denied, but my configuration is correct
there i have set permissions right, i am 99% sure about this.

can this relate also the problem that times between central manager and
solaris have so big difference (maybe days), havent checked recently and
cant remember... ( i could set the solaris to update time from central
manager using ntp)

THX all for any kind off answer.

( i will also try to share my knolidge about condor (3 months using it)
when i have more time (get the current test-bed to work) )

______________
> Cheers :D   \
> Antti Eskola \_________________
> Student in software engineering
> Gsm +358 40 772 3302
> (GMT +2)

> Hi,
> I had quite a similar problem I'd reported about to condor-admin. 
> I had an NTP (network time protocol) daemon running in the background
> and changing time. So during the startd invocation it had one time
> stamp, but then, when the time change happened due to NTP, startd was
> stopping sending its classads. Tuning NTP to start before Condor in the
> startup scripts solved the problem. 
> Mark
> On Fri, 2003-11-14 at 19:43, Heinz, Michael William wrote:
> > Zach,
> > 
> > 	Thanks for the response - I am seeing "invalid session" messages in
> > the collector log.
> > 
> > 	I'll try your suggestion, but I am, in fact, using Condor 6.5.5, so
> > it looks like that problem hasn't been fixed after all. Do I need
> to put the
> > fix in all condor configs or just on the machine running the collector?
> > 
> > -----Original Message-----
> > From: Zachary Miller [mailto:zmiller@xxxxxxxxxxx] 
> > Sent: Friday, November 14, 2003 12:29 PM
> > To: condor-users@xxxxxxxxxxx
> > Subject: Re: [condor-users] Windows Condor disappearing/reappearing 
> on grid
> > 
> > 
> > On Fri, Nov 14, 2003 at 12:13:01PM -0500, Heinz, Michael William wrote:
> > > No one ever responded to this problem, so I'm asking again:
> > > 
> > > I seem to be in a situation where machines disappear off the grid 
> even 
> > > though they are both idle and continuing to operate. Rebooting each 
> > > machine in question seems to relieve the problem. More strangely, it 
> > > sometimes seems as if simply logging into the machine and doing a 
> > > "condor_status" and waiting a few minutes is all that's needed.
> > > 
> > > Does anyone have any suggestions?
> > 
> > it looks like you are running into a bug with sessions expiring.  i'm
> > guessing that if you look in your CollectorLog, you'll see a bunch
> of lines
> > saying "attempt to open invalid session".
> > 
> > the bug was fixed in the 6.5 series.
> > 
> > if you'd like to work around it, you can turn off sessions entirely 
> if you
> > are not using KERBEROS, X509, 3DES, BLOWFISH, MD5, or any other security
> > features.  simply add:
> > 
> >   SEC_DEFAULT_NEGOTIATION = NEVER
> > 
> > to your condor_config file.  let me know if you continue to have
> troubles.
> > 
> > 
> > cheers,
> > -zach
> > 
> > Condor Support Information:
> http://www.cs.wisc.edu/condor/condor-support/
> > To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with unsubscribe
> > condor-users <your_email_address>
> > 
> > 
> > Condor Support Information:
> > http://www.cs.wisc.edu/condor/condor-support/
> > To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
> > unsubscribe condor-users <your_email_address>
> 
> Condor Support Information:
> http://www.cs.wisc.edu/condor/condor-support/
> To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
> unsubscribe condor-users <your_email_address>
> 


Condor Support Information:
http://www.cs.wisc.edu/condor/condor-support/
To Unsubscribe, send mail to majordomo@xxxxxxxxxxx with
unsubscribe condor-users <your_email_address>